• Blog
  • Insurance & Banking: AI QA & Observability -  Supporting SDLC

    Sedona Digital

    March 13, 2026

    This article is written and published in conjunction with Sedona Digital

    By some margin, the most common concerns we hear from insurance and banking CTOs right now can be paraphrased to:
     
    • "We have 50 new AI apps in dev, and only 5 in prod. And TBH, the gap is widening."
    • "I'm not sure my QA/testing is meeting quality or compliance standards for AI generated app code yet, and certainly not at the pace we're producing it."
    This used to crop up once a month, now it's once a week. From big firms with City offices and hundreds of devs and testers, to specialist FS providers from Merseyside to Kent running only a couple of scrum teams. There is no enterprise FS-ready COTS product, no manual, framework nor architecture on the market that solves this and, IMHO, anyone that says otherwise is definitely selling something.
     
    The traditional gap at the intersection of IT and the business is widening because of the cultural change implications of AI - not shrinking just because we can gen business app code at many times the rate of 12 months ago. No longer just the usual business process automation concerns, or chasing the fast-moving and elusive understanding of the art of the technically possible. Not even the relatively recent but now well-described transparency concerns of agentic decision making, especially in regulated sectors. Now business expectations of IT AI, fuelled by their non-tecchies' new abilities to describe and vibe something cool (if perhaps not secure, compliant or scalable), is stretching the gap to a gulf.
     
    These are some of the reasons that the dev-to-prod debt pain is worsening, and often we're feeling it most in QA, compliance, security and risk. Unfairly seen as the 'No-Men', entrenched on the other side of the gen AI revolution, to where our best developers often aren't interested in building bridges.
    A CTO said to me this week "I love code-gen AI, it makes me feel like a kid again". As parents of young children, we instantly thought about the good habits and guardrails our kids needed to thrive and grow safely, and how much we had to learn to instil good habits and processes today (and there was never one tool for this job either).
     
    At Sedona our customers are trusting us to lead a collaborative approach which helps to close the gaps between gen AI dev, test and prod. We lead, interview and scope the 'observability consumer' needs that are key to how your business operates (business compliance? risk? product/service management?) speaking the language of senior FS corp functions leads. We then use this to inform a bespoke meta-data design for observability which your devs adopt as standard app outputs, and embed quickly into your tests/QA. We build the data management layer for this using your preferred cloud data tech, and we then build the new observability analytical insights needed by your business and IT QA teams.
     
    It takes a few months, but it's a strategic approach to part of the challenge above which can make a big difference to you moving apps into prod with speed and confidence. At the same time, helping to alleviate the bottleneck on QA by proactively producing the code and run observability data your business needs.
     
    If you're interested, drop me a line and we can chat through our latest work in this area and how your business could benefit from a support of tailored, strategic approach.


    Tim Bull - Director Enterprise Solutions, Sedona Digital

    Tim spends much of his time at Sedona working with UK FS leaders, driving SDLC and AI agent operations modernisation including QA, observability, requirements and backlog management. He tells his kids that he's "spying on the AI robots to keep a lookout for early Terminators".

    Get in touch